Web Posted on: February 24, 1998
ADAPTING WEB BROWSERS FOR ACCESSIBILITY
Paul Hendrix
Center for Accessible Technology
2547 8th Street, 12A, Berkeley, CA 94710
(510) 841-3224
cforat@earthlink.net
http://www.el.net/CAT
Michael Birkmire
Learning Independence Through Computers, Inc.
1001 Eastern Avenue 3rd floor, Baltimore MD 21202
(410) 659-5462
mikebirkmire@earthlink.net
http://www.linc.org
This article provides an overview of ways to make Web browsers accessible, so that through these programs, users can gain access to the information on the Internet. Our focus is on general techniques for adapting major commercial browsers. A more comprehensive treatment of this subject, including specific information on adaptive products and their manufacturers, can be found at the Web sites referenced above.
1. ACCESS ISSUES
While a person's home computer can be modified for access, the data on the Web is not under the user's control, and may not be presented in an accessible manner; text may be displayed as text or a graphic, links may appear anywhere on a page, or be integrated into a graphic, and so on. Moreover, many Web pages are specifically designed to take advantage of the display capabilities of major commercial browsers, which makes these programs almost required interfaces for accessing the Web. Although there are some special purpose browsers which provide superior access for people with certain disabilities, for many, effective access to the Internet means effective access to one of two pieces of commercial software: Netscape, and Microsoft's Internet Explorer.
As usual, the access issues break down into those affecting the user's ability to input information into the computer, processing aids to simplify or accelerate the operation of the computer, and issues affecting output of information from the computer.
2. INPUT
Generally, the approach to modifying input for a particular computer task involves two components: determining the functions of the software the user wishes to utilize, so that the task can be broken down into its necessary functions, and examining the software to identify the corresponding command features, how they are activated, and how they can be modified.
A. MOUSE ACCESS
Most aspects of Web browsing can be easily accomplished without a keyboard. Because links can appear anywhere on a Web page, they are most conveniently accessed by clicking with a mouse or alternative pointing device. Additionally, most browsers have toolbar icons and buttons, and at least menu options, to allow everything to be controlled by mouse. As a consequence, people who can use the mouse (or an alternative pointing device) can operate Web browsers fairly easily.
When browsing the Web, the need to enter text is infrequent. A URL (the unique Internet address of each Web page) may need to be entered, although pages are more frequently accessed by following hypertext links, and, once accessed, can be bookmarked so that the URL need not be typed again. Text may also need to be entered in a search engine. In neither case, however, is the amount of text very significant; occasional use of an onscreen keyboard would provide fairly complete access to a browser for someone who could use a mouse or mouse emulator.
Alternative Pointing Devices: There are a number of alternatives to the standard mouse. Some of these are readily available from computer supply stores; these options include trackballs, joysticks, and touch-sensitive trackpads that respond to the movement of a finger over their surface. There are also special mice and trackballs that have been modified to use an external switch to activate the mouse click.
Some more esoteric hands-free options allow movement of the user's head or eyes to control the onscreen cursor. Mouse emulation can also be achieved with switch-controlled software. For example, one type causes a horizontal line to move down the screen from top to bottom; a second switch activation causes the cursor to move along the horizontal line from left to right, and a third stops it at the desired point on the screen
OnScreen Keyboards: Entering text without a keyboard, using a mouse or mouse alternative, requires an onscreen keyboard. These are graphical representations of the keyboard that are displayed on the computer monitor. Moving the mouse over one of the onscreen keyboard "keys", and either clicking or "dwelling" on the key for a preset period of time will cause that key to be typed. Customizable onscreen keyboards can allow the user to type words or multi-letter commands by activating a single "key"; this allows the creation of custom keyboard layouts specifically designed for the control of Web browsers.
B. KEYBOARD ACCESS
A more complex problem involves controlling a Web browser without a mouse. In many cases, the key to making standard commercial software accessible is determining how to operate it entirely by keyboard commands. If this can be accomplished, any access method that will emulate a keyboard entry can be used: an alternative keyboard layout, a custom switch scanning matrix, Morse Code, speech macros for a voice input system. Accordingly, a crucial physical access issue is the keyboard accessibility of Web browsers, and how we can then make them accessible to switch and alternate keyboard users.
Keyboard Access to Browsers: First, it is helpful to know how to operate the system itself from the keyboard. In addition to information contained in manuals and in Help files, there is a comprehensive list of Windows 95 keyboard shortcuts on-line, and books addressing such issues for the Windows 3.1 system. The Macintosh operating system, originally designed for mouse access, does not incorporate a comprehensive set of keyboard shortcuts, but the Mac can be given this functionality with utility programs that allows dialog boxes, menu bar items, and on-screen hot spots to be accessed with keyboard commands.
Essential Browser Functions: Obviously, what's "essential" depends on the user's needs. Moving forward and back between pages, creating and accessing bookmarks to certain pages, scrolling up and down a displayed page, and selecting and activating links to other pages are minimum functions, and would be a good starting point for a child or new user, for example. More sophisticated users, or those wanting to do on-line research, would need more functions, especially access to a search engine.
On the Macintosh, a keyboard utility that provides access to menu items can overcome most physical access barriers except the biggest one: there is no way to select and activate links on the Macintosh with using a mouse or mouse emulation.
On Windows-based machines, keyboard access to the menus is built-in. However, the crucial advantage of the Windows versions of Internet Explorer and Netscape for Windows, is that they allow links to be selected by pressing the Tab key, making the entire program keyboard accessible.
A comprehensive list of the keyboard commands necessary to operate the various releases of Navigator and Internet Explorer, and a list of suggested key browser functions, can be found on the authors' Web pages, referenced above.
Alternative Keyboards: There are several programmable alternative keyboards available, which when used with related, allow the user to design custom keyboard layouts that can include special keys for web browser commands. If some thought is given to the browser functions for which access is desired, a specialized keyboard for Web navigation can be created. Some samples can be found on the authors' Web pages.
C. SWITCH ACCESS
Morse Code: Despite its archaic sound, Morse Code is actually one of the simplest and fastest means of controlling the computer without a keyboard. One or two switches are used to enter the dots and dashes of Morse Code; the computer interprets these codes as keyboard keys. Any switches can be used, including switches activated by the user's breath, or eye blinks. Of course, the user must learn Morse Code, but this can be accomplished with a week or so of practice, and fairly high input speeds can be reached. However, it is most effective for keyboard replacement; mouse emulation with Morse Code is clumsy ("mouse up", "stop", "mouse right").
Switch Scanning: Switch scanning is the least physically demanding means of computer input: all the user has to do is activate a switch at the right time; it is also very slow. The user is presented with a grid of options, the computer scans through these options in a certain pattern, and when the desired item is reached, the user activates the switch. Often, the program will scan through the grid or matrix of items row by row; when the appropriate row is reached, the user activates the switch and that row is scanned, item by item, until the user activates the switch again. As with Morse Code, this system works best with keyboard commands; mouse emulation with switch scanning is fairly cumbersome.
Some scanning programs allow the creation of special scanning grids or matrices that can include specialized browser commands. This can be a much more efficient means of controlling a web browser than switch selecting the individual keyboard keys. The program may allow the user to select commands that also cause the display of different matrices for different functions. This avoids the necessity of having all functions contained in a single matrix and therefore reduces the time needed to scan within a given matrix
D. VOICE INPUT
Most voice input systems are not very good at controlling the mouse; it tends to be a pretty cumbersome process ("mouse up", "stop", "mouse left"). Therefore, keyboard based Web access methods will be helpful in using voice input as well. Once the keyboard command to control a browser is known, a speech macro to do this by voice can be created.
The fully featured versions of the major voice input will allow voice control of Web browsers. Browsers that are entirely controllable by keyboard commands, can be easily controlled by voice; those that cannot will require the use of the voice input system's mouse emulation features, which can be cumbersome ("move mouse up", "move mouse left"). There are also voice input applications designed solely to control the Netscape Navigator Web browser; these programs allow you to activate links on a Web page simply by speaking the name of the link..
3. PROCESSING AIDS
For our purposes, these fall into three categories: word prediction/abbreviation expansion programs, spell checkers, and text-to-speech software. Word prediction and abbreviation expansion programs work with Web browsers the same way they work with other software. These programs provide means of accelerating the user's input by minimizing the number of keystrokes or selections necessary to type text. Such programs are especially useful for people with laborious input methods, such as switch scanning, but are of primary importance for e-mail and on-line chat rooms, where large amounts of text must be entered. The same is true for spell checkers; the text that must be entered while actually browsing the web is usually limited to one or two word search terms.
Text-to-speech programs that speak highlighted text can be helpful to people with learning disabilities or low vision in accessing text on Web pages. The primary difficulty is when words on a page are presented as graphics, rather than actual text; the text-to-speech programs cannot read this, nor can the screen readers relied upon by blind users.
Two specialized browsers may also be helpful for people who need text-to-speech help in accessing the Web. PwWebSpeak, a Web browser designed for blind users, displays the text from a Web page in a de-columnized form and reads it aloud. It does not display the graphics, but it does highlight the text being read. Ultimate Reader can be configured to work like a browser; it will access Web pages, reformat to display their text component in a simple manner, and read the text aloud, highlighting as it goes.
4. OUTPUT
Adapting the output of a computer - which is primarily comprised of the information displayed on the monitor screen - can take the form of modifying this visual output, or supplementing or replacing it with auditory or tactile output.
Modifying the visual output usually means enlarging the text, or the cursor, or the entire displayed image. Enlarging the text can be accomplished within the preferences controls of the browsers themselves. The Windows 95 operating system has built-in provisions for enlarging the cursor; there are also several utility programs, for both platforms, to accomplish this. Enlargement of the entire displayed image can be accomplished with a screen magnification program. Screen magnification programs can be difficult to use, however, since only a part of the enlarged image can be displayed on the monitor at a time; it can be very hard to keep track of your place. The visual image can also be modified by altering the color of the text and background for better contrast. Most Web browsers allow the user to control this feature, most screen magnification programs also incorporate color control, and Windows 95 has a high-contrast display setting built into its accessibility control panel.
By supplementing the visual output, we are referring to having the computer read aloud text that is visually identified and selected. This can be accomplished with several text-to-speech programs, described in the Processing Aids section, above.
Replacing the visual output with auditory output refers to using a screenreader program to audibly interpret the graphical information sent to the display screen by the computer or to display the information on an electronic refreshable Braille display device which is read as if it were conventional paper Braille. There are a number of these programs available some of which have the capability of driving a Braille display, all of which deal somewhat differently with the problem of presenting the unpredictable graphical data produced by the Web browsers in audible or tactile form.
Screen readers worked well under a text-based, command line operating system like DOS. The job of audibly representing the two-dimensional, simultaneously displayed, graphical information in the Mac and Windows environment is a complicated one, and users require considerable training to use these products effectively. On Windows-based machines, they also require a speech synthesizer to produce sound output. There are a number of external synthesizer devices available; many screenreaders now support software-based text-to-speech engines that take advantage of Windows sound cards.
The process of accessing the Web with a screen reader is somewhat circular. The original HTML document is in text form; the Web browser reads this text document and converts it into a graphical representation of the Web page, and the screen reader attempts to audibly translate this representation as it appears on the screen. One specialized browser, PwWebSpeak, audibly interprets the HTML code directly.